Why Interrogation Contamination Occurs

Why Interrogation Contamination Occurs Richard A. Leo* I. INTRODUCTION On July 25, 1983, George Allen was convicted of the capital murder and rape of Mary Bell.' St. Louis Police had initially suspected Kirk Eaton, a convicted sex offender who had recently been released from prison. Kirk Eaton had been spotted by a police officer near Ms. Bell's apartment complex, and had disappeared shortly after Ms. Bell was murdered. More than a month later, police arrested George Allen, a diagnosed schizophrenic, mistakenly believing him to be Kirk Eaton. Although police discovered at the police station that Mr. Allen was not Kirk Eaton, a sex crimes unit officer nevertheless interrogated Mr. Allen about a host of unrelated sex crimes in the area. Homicide Detective Herbert Riley then interrogated Mr. Allen about the Bell murder. According to Detective Riley, he questioned Mr. Allen for fifteen to twenty minutes prior to showing him photographs of Ms. Bell's apartment, then advised Mr. Allen of his Miranda rights, and thereafter turned on a cassette recorder, during which Mr. Allen confessed to the rape and murder of Mary Bell. The confession was the sole basis for charging Allen with these crimes, and, along with vague serological evidence that was eventually shown to be false, ultimately led to his conviction. There was no physical evidence linking Allen to the crime or supporting the confession. DNA testing in 2003 and 2010 conclusively eliminated Mr. Allen as the source of semen left in Ms. Bell, on the * Visiting Professor of Law, University of California, Los Angeles (2013-2014) and Professor of Law and Dean's Circle Research Scholar, University of San Francisco. I thank Olga Axselrod, Joshua Dressler, Rosa Greenbaum, and Carrie Leonetti for helpful comments. 1 Unless otherwise noted, the facts of George Allen's case are drawn from the following sources: State v. Allen, 684 S.W.2d 417, 419-21 (Mo. Ct. App. 1984); State ex rel Koster v. Green, 388 S.W.3d 603, 606-16 (Mo. Ct. App. 2012); Writ of Habeas Corpus, In re George Allen Jr. v. David Dormire (2012) (No. 1 AC-CC00634); Mem. in Supp. of Pet. for Writ of Habeas Corpus, In re George Allen Jr. v. David Dormire (on file with author); First Am. Pet. for Writ of Habeas Corpus and Mem. in Supp., In re George Allen Jr. v. David Dormire, No. I 1AC-CC00634, (2012); Aff. of Richard A. Leo, Ph.D., J.D., In re George Allen Jr. v. David Dormire (2012) (on file with author); Boaz Sangero & Mordechai Halpert, Proposal to Reverse the View of a Confession: From Key Evidence Requiring Corroboration to Corroboration for Key Evidence, 44 U. MICH. J.L. REFORM 511, 533-37 (2011); Maurice Possley, George Allen, Jr., NAT'L REGISTRY OF EXONERATIONS (Mar. 5, 2013), http://www.law.umich.edu/special/exoneration/pages/casedetail.aspx?caseid-4091. Most of these documents, as well as many others, can be accessed at http://www.innocenceproject.org/content/legal-documents-fromallen-vdormire.php (last accessed on Feb. 19, 2013). 193

194 OHIO STATE JOURNAL OF CRIMTNAL LAW [Vol 11: 1 towel in which the murder weapon was wrapped, and elsewhere at the crime scene. On November 14, 2012-after nearly thirty years in prison-missouri Circuit Judge Daniel Green vacated Allen's conviction, ruling that the State of Missouri had withheld exculpatory evidence excluding Mr. Allen and pointing to someone else as the perpetrator. 2 Mr. Allen's confession contained numerous factual errors about the crime, including the time of day it had occurred, the appearance of Ms. Bell, what she was wearing, where the weapon had been obtained and left behind, and the manner in which she had been killed. Yet the prosecutor argued at trial that the accuracy of the confession was corroborated by two non-public details revealed by Mr. Allen that were supposedly known only by the true perpetrator, and Detective Riley testified that he learned both facts only after Mr. Allen had first mentioned them in his confession. Although George Allen's confession statement did not, in fact, mention any verifiable non-public facts known only to the true perpetrator, it did contain numerous non-public crime scene facts that appeared to cohere into a detailed and persuasive narrative corroborating his guilt. Here is how the Missouri Court of Appeals summarized the content of Mr. Allen's confession when it reviewed his case in 1984: Defendant confessed to the murder of Mary Bell in a taped statement given to the homicide detective. In defendant's taped statement to the detective, he acknowledged that he had been at Mary's apartment complex in February, at the time of the "big snow," the day of her murder. He knocked on her door and when she opened it, forced his way in. He followed her up the stairs where she tried to hold the bedroom door shut against him. He remembered chasing her through the kitchen, also on the second floor, across from the bedroom. In the kitchen she grabbed a butcher knife, about 12 inches long. While grappling with her, he knocked it from her hand. He admitted she was stabbed during the fight and that he had sex with her on her brass bed or on the floor. While in the apartment, he heard knocks and then banging on the front door and a woman's voice calling out "Sherry" or "something like that." He thought the knocking on the door was from next door when he heard the next door neighbor open and close her door. He described Mary as white 2 The semen, which was attributed to George Allen at trial, turned out to be from Ms. Bell's boyfriend. But it was then discovered that there were additional semen stains left behind by someone else but unfortunately consumed by serological testing at trial that excluded George Allen and Ms. Bell's boyfriend, and those results were hidden by police. See Writ of Habeas Corpus, In re George Allen Jr. v. David Dormire, No. 1 IAC-CC00634, (2012). The police contaminated two witnesses after the fact in order to get corroboration of these details, one of whom was subjected to hypnosis to get her to recall the facts that George Allen had allegedly offered up in the confession. Both witnesses were found to be unreliable. Id.

2013 WHY INTERROGATION CONTAMNATION OCCURS 195 with dark hair, about 20 to 25 years old, with large breasts. Then, when shown a picture of Mary Bell, he admitted she was the woman he had stabbed and killed. 4 If Mr. Allen's confession statement is demonstrably false-as we now know-where did all these crime scene details come from? In his trial testimony, Detective Riley emphasized that he did not provide any case facts to Mr. Allen, or even discuss the crime with him, prior to turning the tape recording on. The partially recorded interrogation transcript, however, reveals the opposite. Detective Riley repeatedly leaked details of the crime to Mr. Allen; fed Mr. Allen case facts through leading and highly suggestive questions; pressured Mr. Allen to adopt the desired answer when he could not provide it; and corrected Mr. Allen when he provided incorrect answers, which happened frequently. In addition, Detective Riley showed Mr. Allen photographs of the outside and inside of the apartment complex (including the bedroom), the crime scene, the murder weapon, and the murder victim. In other words, Detective Riley contaminated George Allen's confession. The problem of police contamination is pervasive in documented false confessions, as Brandon Garrett's recent empirical research has shown.6 Specifically, Professor Garrett examined the first 250 post-conviction DNA exonerations of innocent prisoners since 1989,' forty of whom had falsely confessed to rapes and murders. In thirty-six of the thirty-eight cases (97%), the confessions contained a litany of specific details about how the crime occurred. As Garrett puts it: "All of but two of the forty exonerees studied told police much more than just 'I did it.' Instead, police said that these innocent people gave rich, detailed, and accurate information about the crime, including what police described 4 Allen, 684 S.W.2d at 420. One contamination strategy used by Detective Riley was to repeatedly ask Mr. Allen if he was sure of an answer if Mr. Allen provided the wrong answer. Another contamination strategy was to mischaracterize what Mr. Allen had previously said by falsely summarizing it and including crime facts or statements that Mr. Allen did not, in fact, make, but the most striking contamination strategy that Detective Riley used was to show Mr. Allen photographs of (1) the outside of the apartment complex; (2) the inside the apartment complex, including the bedroom; (3) the crime scene; (4) the murder weapon; and (5) the murder victim. All of these contamination strategies had the effect of educating Mr. Allen about where, when, how and why the crime occurred. They allowed Detective Riley to falsely present Mr. Allen's confession as if it contained unique and non-public crime details when, in fact, it contained nothing more than Mr. Allen parroting back Detective Riley's suggestions or inferring the correct, or a seemingly correct, answer from what Detective Riley had been repeatedly, and aggressively, suggesting to him. 6 Brandon L. Garrett, The Substance offalse Confessions, 62 STAN. L. REv. 1051 (2010). Id. at 1052-53 n.2. Id. at 1054.

196 OHIO STATE JOURNAL OF CRIMINAL LAW [Vol 11: I as 'inside information' that only the true culprit could have known." 9 This included facts that matched the crime scene evidence, scientific evidence, expert evidence and/or accounts by the victim.' 0 Yet all of these individuals were factually innocent: not only had they not committed the crimes of which they were convicted, but none of them had been present at the crime scene or knew any of the non-public details memorialized in their confession statements." Instead, police investigators disclosed non-public crime details to these suspects during their (partially or entirely unrecorded) interrogations. They did so by telling the suspects how the crime happened and feeding them crime facts.1 2 The contamination included showing defendants police reports, co-defendant statements, crime photographs and even taking them to the crime scene itself.' 3 George Allen's case was not part of Garrett's data set, but it illustrates the problem and pattern of interrogation contamination that Garrett's study describes so well. Allen's case also illustrates how police interrogation contamination increases the risk that a demonstrably false confession will nevertheless be perceived by criminal justice officials and juries as true and thus lead to an erroneous conviction. When police interrogators disclose non-public facts to suspects, they often pressure suspects to incorporate the contaminated information into their post-admission narrative. The contaminated details contained in false confessions have been referred to as misleading specialized knowledge.1 4 The presence of misleading specialized knowledge in contaminated false confessions creates the illusion that the suspect volunteered detailed inside information about the crime that "only the true perpetrator could have known." Empirical research has shown that a cue people rely on when evaluating the accuracy of confession evidence is the richness of detail in the confession statement, especially if it contains details that were not made public.' Yet contaminated false confessions appear every bit as rich and detailed as true confessions, and controlled studies have shown that people are not good at discriminating between the two.1 6 The risk of wrongful conviction is further compounded when police and prosecutors erroneously assert in testimony and argument at trial-as occurred in George Allen's case-that the non-public details contained in the defendant's 9 BRANDON L. GARRETT, CONVICTING THE INNOCENT: WHERE CRIMINAL PROSECUTIONS Go WRONG 19 (2011). 10 Garrett, supra note 6, at 1057. " Id. at 1054. 12 Id. at 1070-71. 13 Id. 14 See RICHARD A. LEO, POLICE INTERROGATION AND AMERICAN JUSTICE 254-55 (2008). is DAN SIMON, IN DOUBT: THE PSYCHOLOGY OF THE CRIMINAL JUSTICE PROCESS 162 (2012). 16 Saul M. Kassin et al., "I'd Know a False Confession ifi Saw One": A Comparative Study of College Students and Police Investigators, 29 LAw & HUM. BEHAV. 211 (2005).

2013 WHY INTERROGATION CONTAMINATION OCCURS 197 (contaminated) confession originated with the suspect, not with the interrogating detectives. The problem of police interrogation contamination is well-documented but not well-understood. While there is extensive empirical social psychological research literature explaining why police interrogators sometimes elicit false confessions from innocent suspects,' 7 there is no comparable empirical research analyzing why police interrogators sometimes contaminate the narratives of innocent false confessors. Yet contamination corrupts the integrity of confession evidence and may be the primary reason why false confessions so often lead to wrongful convictions when introduced into evidence at trial. 18 Numerous scholars have documented and analyzed the phenomenon of interrogation contamination and its consequences,1 9 but no scholar has really explained why American police interrogation leads to the (presumably inadvertent) leakage and disclosure of nonpublic crime facts in so many false confession cases. In this short article, I seek a deeper understanding of police interrogation contamination than currently exists in the empirical and legal literature. I analyze not whether, but why police interrogation contamination occurs. Police interrogation contamination seemingly presents a puzzle since it is openly recognized as improper by police, who universally condemn it in their training programs and publications.2o To better understand why police detectives appear so often to engage in this prohibited practice when interrogating the innocent, we must first examine the psychological dynamics of the interrogation process as a whole. I argue that contamination is a predictable consequence of the American method of guilt-presumptive accusatory interrogation and the confirmation biases it promotes, reinforces, and perpetuates. Police interrogation contamination is not so much a problem of individual or poorly trained detectives as it is a systemic feature of the adversarial assumptions, goals and logic of mainstream American police interrogation practice and culture. I further argue that contamination may 17 For a recent review, see Saul M. Kassin et al., Police-Induced Confessions: Risk Factors and Recommendations, 34 LAw & HUM. BEHAV. 3 (2010). 18 LEO, supra note 14, at 253-66. 1 See, e.g., GISLI H. GUDJONSSON, THE PSYCHOLOGY OF INTERROGATIONS, CONFESSIONS AND TESTIMONY 316-20 (1992); Richard J. Ofshe & Richard A. Leo, The Decision to Confess Falsely: Rational Choice and Irrational Action, 74 DENV. U. L. REV. 979 (1997); Richard A. Leo & Richard J. Ofshe, The Consequences offalse Confessions: Deprivations ofliberty and Miscarriages ofjustice in the Age of Psychological Interrogation, 88 J. CRIM. L. & CRIMINOLOGY 429 (1998); Saul M. Kassin, A Critical Appraisal of Modern Police Interrogations, in INVESTIGATIVE INTERVIEWING: RIGHTS, RESEARCH AND REGULATION 207 (Tom Williamson ed., 2006) [hereinafter INTERVIEWING]; LEo, supra note 14, at 165-94; Sara C. Appleby et al., Police-Induced Confessions: An Empirical Analysis of Their Content and Impact, 19 PSYCH. CRIME & L. 111 (2011). 20 FRED E. INBAU ET AL., CRIMINAL INTERROGATION AND CONFESSIONS 369-71 (4th ed. 2001). See also Joseph P. Buckley, The Reid Technique ofinterviewing and Interrogation, in INVESTIGATIVE INTERVIEWING: RIGHTS, RESEARCH AND REGULATION, supra note 19, at 190.

198 OHIO STATE JOURNAL OF CRIMNAL LAW [Vol 11: 1 be virtually inevitable when investigators mistakenly interrogate the innocent because American police interrogation has no internal corrective mechanism to catch or reverse investigators' pre-interrogation classification errors or their confirmatory, information-conveying interrogation techniques. II. THE CONTEXT AND PSYCHOLOGY OF AMERICAN POLICE INTERROGATION American police interrogators are committed to the goal of convicting the suspects they interrogate. Contrary to the myths of American criminal justice, police interrogators are not neutral or independent fact-finders pursuing a valuefree search for the truth. Rather, once they have judged a custodial suspect guilty, they steadfastly seek to incriminate him. Detectives often see the interrogation process as their best opportunity to build a case against a suspect by eliciting and constructing admissions-ideally a full confession-in order to ensure his conviction. Once American police have decided to interrogate a suspect, an unmistakable conviction psychology underlies and animates their assumptions, methods, strategies and goals. Dan Simon has referred to this tendency more broadly as the adversarial pull. 21 Interrogation is guilt-presumptive. 22 Detectives are trained to interrogate only those individuals whom they have first reasonably concluded are guilty of the 23 crime. Interrogation is thus not supposed to be exploratory or the starting point of a criminal investigation, but commence only after police have first conducted a reasonable investigation. 24 "Investigate before you interrogate" 25 is a mantra of current police training manuals. Police are trained to determine whether a suspect is guilty in the investigation that precedes interrogation, and thus any subsequent interrogation is necessarily founded on a belief that the suspect participated in and/or committed the crime in question. A logical corollary is that the interrogation is not a process designed to separate the innocent from the guilty, determine if the suspect committed the crime, evaluate his or her denials, or even discern the truth. 26 21 SIMON, supra note 15, at 31-33. 22 INBAU ET AL., supra note 20, at 8. See also Saul M. Kassin et al., Behavioral Confirmation in the Interrogation Room: On the Dangers of Presuming Guilt, 27 LAW & HUM. BEHAV. 187 (2003); LEO supra note 14, at I 10-11. 23 INBAU ET AL., supra note 20, at 8. 24 Id. at 23. 25 Id. 26 In more recent editions of their interrogation training manual, Inbau et al. have claimed that the goal of their guilt-presumptive accusatory interrogation methods is "to learn the truth." Id., at 229. But as others have pointed out, this claim simply does not square with the assumption, psychology and logic of the Reid Method of interrogation and thus cannot be true. See DAVID A. HARRIS, FAILED EVIDENCE: WHY LAW ENFORCEMENT RESISTS SCIENCE 41 (2012). The goal of

2013 WHY INTERROGATION CONTAMNATION OCCURS 199 The goal of interrogation is to move the presumed guilty suspect from denial to admission, ideally a full narrative confession.27 To accomplish this goal, police detectives employ interrogation methods that rely on pressure and persuasion. Although there are hundreds of techniques, the basic psychology of American police interrogation involves a few basic overarching methods and strategies that usually occur in predictable and repetitive sequences. 28 The most fundamental technique is the accusation, which usually triggers the start of the actual interrogation. Detectives typically accuse suspects, often repeatedly, of participating in or committing the crime and of lying. In response to accusations, most suspects deny knowledge of and/or participation in the crime. Anticipating this, police interrogators aggressively challenge and attack the suspect's denials, seeking to reverse them. Police detectives view overcoming a suspect's denials as essential to successful interrogation because a suspect who is allowed to persist in denial is not likely to make or agree to an admission. Police interrogators typically attack a suspect's denials by suggesting they are implausible or unbelievable, illogical or impossible, and/or inconsistent with or contradicted by the case facts as the detective knows them. As with accusations, attacks on denials are repeated often throughout interrogation. In addition to accusations and attacks on denials, interrogators also rely on what are known as "evidence ploys" to move the suspect from denial to admission. 29 An evidence ploy is any attempt to make the suspect believe that interrogators possess incriminating evidence against him. 30 For example, an interrogator may confront a suspect with eyewitness identifications, accomplice statements, surveillance video, DNA results, and the like. Evidence ploys may either be true or false: if interrogators possess actual evidence against a suspect, they will confront him with it as they press him to confess; if interrogators do not possess any actual evidence against a suspect, they will often pretend that they do and confront him with the same invented evidence. 31 Like accusations and attacking denials, evidence ploys are one of the most fundamental and common techniques of modern psychological interrogation. 32 Evidence ploys are central to the psychology of interrogation because they are part of a general strategy of communicating omniscience, exposure and capture, and thus the inevitability of conviction. Police interrogators often pretend that they know all the facts of the case but just need to hear what happened from the suspect American Reid-style interrogation is not to learn the truth but to confirm the interrogator's preinterrogation theory of the truth. See LEO, supra note 14. 27 This discussion draws on LEO, supra note 14, at 119-64. 28 Id. at 119-62. 29 Id. at 139-148. 30 Id. ' Id. 32 id.

200 OHIO STATE JOURNAL OF CRIMINAL LAW [Vol 11: 1 (e.g., in order to test his truthfulness), that they know more than the suspect knows or thinks he knows, or that they possess evidence against the suspect of which he is unaware. Interrogators commonly tell a suspect that the evidence against him has been established so irrefutably that the only issue left to discuss is not whether he committed the crime, but why. In this sense, evidence ploys could also be called omniscience ploys, i.e. representations that because the interrogator knows everything about the crime facts, the suspect can no longer possibly or plausibly deny his guilt. Evidence ploys communicate that the suspect has been captured, his guilt has been exposed, and, in effect, he has no meaningful choice but to stop denying and start admitting his knowledge of and participation in the crime. Accusations, attacks on denials, and evidence ploys are designed to break down the suspect's self-confidence and overcome his resistance. 34 Elsewhere I have referred to these as negative incentives. 5 To move a suspect from denial to admission, however, interrogators also rely on what I have called positive incentives,36 namely inducements to motivate a suspect to admit his guilt. While interrogators use negative incentives to get the suspect to perceive that his denials are futile and that his conviction is inevitable, they use positive incentives to persuade the suspect that he will obtain some moral, psychological, material or legal benefit or reward (and avoid a corresponding harm) if he makes or agrees to an incriminating statement. For example, interrogators may appeal to the suspect's conscience, pointing out the moral or religious benefits of confessing; or they may emphasize how they can help portray the suspect's actions more favorably in their reports and court testimony if he cooperates and confesses; or they may imply or suggest that the suspect will receive more favorable treatment from the prosecutor, judge or jury if he stops denying and starts admitting his guilt, sometimes even spelling out the proposed benefits (and corresponding harms). To induce the suspect to agree or admit to the underlying act, interrogators often suggest scenarios of how and why they could have committed the crime. Known as "themes," these scenarios seek to rationalize the suspect's actions by morally, psychologically, or legally downplaying, excusing or justifying the underlying act of the crime. Interrogators advance scenarios to persuade a suspect that if he admits to the act he will-with the interrogator's help-benefit by being able to frame how his act will be understood by prosecutors, judges, juries, and others. The goal is to lead the suspect to believe that if he admits to the underlying act, he can explain his motive in a way that will portray him in the most sympathetic light and minimize his social, moral and legal culpability. 33 Id. 34 Id. at 132-50. 35 Id. at 134-50. 36 Id. at 150-62. 37 id.

2013 WHY INTERROGATION CONTAMNATION OCCURS 201 It is useful for our purposes to distinguish between the pre- and postadmission phases of interrogation. 3 8 As we have seen, in the pre-admission phase, interrogators use accusations, attacks on denial, evidence and omniscience ploys, repetition, inducements, scenarios and other techniques to elicit an "I did it" statement. While investigators' primary goal in the pre-admission phase of the interrogation is to move the suspect from denial to admission, their primary goal in the post-admission phase is to obtain a narrative description of how and why the suspect committed the crime. Just as interrogators are not neutral or impartial in the pre-admission phase of interrogation, they are not neutral or impartial in the post-admission phase either. 39 Instead, detectives sometimes continue to use the same pre-admission techniques-accusations, attacks on denials, evidence ploys, feigned omniscience, etc.-in the post-admission phase if the suspect's responses and account do not match the detective's expectations or theory of the crime. As in the pre-admission phase, the interrogator's primary goal continues to be to incriminate the suspect in order to ensure his successful prosecution and conviction. 40 III. WHY INTERROGATION CONTAMINATION OCCURS Understanding the psychology of interrogation contamination is important for a number of reasons, not the least of which is that police training manuals, classes, and leaders all emphatically advise investigators to avoid contaminating suspects with non-public crime facts. Instead, they recommend that interrogators have suspects independently volunteer these details in order to see whether the suspect possesses personal knowledge about the crime known only to the perpetrator or the police, thus verifying the accuracy of the confession statement. 4 ' If interrogators supply the suspect with non-public crime facts, there is no way of discerning whether his detailed confession is accurate or merely the product of police interrogation contamination. This elementary law enforcement principle is widely accepted in American police work. As the President of the leading interrogation training firm (Reid & Associates) has emphasized, "it is imperative that investigators do not reveal details of the crime so that they can use the disclosure of such information by the suspect as verification of the confession's authenticity." 4 2 In the 2004 edition of their interrogation training manual, the authors of the Reid & Associates interrogation training manual write that: 3 See Ofshe & Leo, supra note 19. 3 See LEO, supra note 14, at 165-94. 40 Id. 41 INBAU ET AL., supra note 20, at 369-71. 42 Buckley, supra note 20, at 204.

202 OHIO STATE JOURNAL OF CRIMNAL LAW [Vol 11: 1 After a suspect has related a general acknowledgment of guilt, the investigator should return to the beginning of the crime and attempt to develop information that can be corroborated by further investigation, and should seek from the suspect full details of the crime and information about the suspect's subsequent activities. What should be sought particularly are facts that would be known only by the guilty person (e.g., information regarding the location of the murder weapon or the stolen goods, the means of entry into the building, the type of accelerant used to start the fire, or the type of clothing on the victim). When developing corroborative information, the investigator must be certain that the details were not somehow revealed to the suspect through the questioning process, news media, or the viewing of crime scene photographs. In this regard, it is suggested that early during an investigation, a decision be made by the lead investigator as to what evidence will be withheld from the public [and press], as well as from all suspects. This information should be documented in writing in the case file so that all investigators are aware of what information will be withheld. 43 Thus, as even police themselves recognize, contamination thwarts the investigative goal of eliciting verifiably accurate and reliable confessions. Why then does it occur? The foundational reason is that the guilt-presumptive psychology of American police interrogation triggers an aggressive set of confirmation biases as investigators use the accusatory techniques of interrogation to elicit a confession that confirms their prejudgment of guilt, leading them to provide details of the crime to the suspect as part of their pre- and post-admission interrogation strategies. Yet American interrogation has no internal mechanism to catch or correct these biases when an innocent suspect is mistakenly targeted for interrogation. Because it is guilt-presumptive, the goal of American police interrogation is not to get the truth, which is presumed, but to get a confession that will assist the prosecution in successfully convicting the defendant. Yet, as we have seen, sometimes police investigators get it wrong, and mistakenly subject an innocent person to an accusatory interrogation process that is designed for the guilty. The "misclassification" error is a predicate for understanding the logic of police interrogation contamination in false confession cases, and all the available evidence suggests that misclassification errors are not uncommon in criminal 43 INBAU ET AL., supra note 20, at 217.

2013 WHYINTERROGATION CONTAMNATION OCCURS 203 investigations generally,"" or in police interrogations specifically. 45 As I will try to show below, misclassification errors beget contamination errors. There are several well-known sources for the misclassification error. 46 The most salient is that American police interrogators are trained to believe that they can reliably infer whether a suspect is lying or telling the truth from his body language, demeanor, mannerisms, gestures, attitudes, styles of speech and other verbal and non-verbal behaviors. 47 The ideology that detectives are, or can become, highly accurate human lie detectors-whether through training or their on-the-job experience-is deeply ingrained in police culture. 48 Yet it has been shown to be false by extensive social science research demonstrating that people are poor at making accurate judgments of truth and deception in general, 49 that the behavior cues police rely on in particular are not diagnostic of deception, 50 and that investigators cannot distinguish truthful from false denials of guilt at rates significantly greater than chance, 5 but instead routinely make confidently held yet erroneous judgments. 5 2 In addition to flawed training and the perseverance of false beliefs in the accuracy of human behavioral lie detection, other sources of the misclassification error include relying on gut hunches, intuitive assumptions, stereotypical thinking, and crime-related schemas or profiles." And, of course, police detectives are also subject to a myriad of normal human decision-making biases that cause people to " See EDWARD CONNORS ET AL., U.S. DEP'T. OF JUSTICE, CONVICTED BY JURIES, EXONERATED BY EVIDENCE: CASE STUDIES IN THE USE OF DNA EVIDENCE TO ESTABLISH INNOCENCE AFTER TRIAL (1996) (reporting that 25% of primary suspects were excluded in the first 10,000 DNA cases tested by the FBI in the 1990s). 4s Richard A. Leo & Steven A. Drizin, The Three Errors: Pathways to False Confession and Wrongful Conviction, in POLICE INTERROGATION AND FALSE CONFESSIONS: CURRENT RESEARCH, PRACTICE, AND POLICY RECOMMENDATIONS 9-30 (G. Daniel Lassiter & Christian A. Meissner eds., 2010); see also SIMON, supra note 15, at 17-49. 46 Leo & Drizin, supra note 45. 47 LEo, supra note 14, at 78-118. 48 id 49 ALDERT VRiu, DETECTING LIES AND DECEIT: PITFALLS AND OPPORTUNITIES (2d ed. 2008); see also Charles F. Bond, Jr. & Bella M. DePaulo, Accuracy ofdeception Judgments, 10 PERS. SOC. PSYCHOL. REV. 214 (2006). so Aldert Vrij et al., An Empirical Test of the Behavior Analysis Interview, 30 LAW & HUM. BEHAV. 329 (2006); see also Saul M. Kassin & Christina T. Fong, "I'm Innocent!": Effects of Training on Judgments of Truth and Deception in the Interrogation Room, 23 LAW & HUM. BEHAV. 499 (1999); Jaume Masip et al., Is the Behavior Analysis Interview Just Common Sense?, 25 APPL. COGNIT. PSYCHOL. 593 (2011). si See Vrij et al., supra note 50. 52 Christian A. Meissner & Saul M. Kassin, "He's Guilty!:" Investigator Bias in Judgments of Truth and Deception, 26 LAW & HUM. BEHAV. 469, 469-80 (2002). 53 See Leo & Drizin, supra note 45.

204 OHIO STATE JOURNAL OF CRIMNAL LAW [Vol 11: 1 mistakenly believe things that are not true. 54 These include the tendency to attribute more meaning to random events than is warranted, to base conclusions on incomplete and unrepresentative information, and to interpret ambiguous evidence to fit one's preconceptions. 55 All of these biases are not only amply present in police work but compounded, if not institutionalized, by the adversarial nature of American police interrogation. 56 Yet for a number of reasons police interrogators are not likely to recognize their misclassification errors. First, detectives tend to doubt the honesty and character of the suspects they encounter, many of whom are repeat offenders. This tendency to perceive interview suspects as guilty has been widely documented, 57 and empirical research has shown it to be exacerbated by mainstream American 58 police interrogation training." Second, like everyone else, police tend to seek, believe and remember information that is consistent with what they already believe. 59 Research across a variety of contexts has demonstrated that once people form beliefs, they search for, interpret, and create subsequent information in ways that verify their existing beliefs, while disregarding contradictory information.o Third, detectives not only assume that they can reliably separate the innocent from the guilty in the investigative process as we have seen, but also that they infrequently make any sorting mistakes. 61 As a result, American police detectives assume that, with rare exceptions, they only interrogate the guilty. The President of Reid & Associates captured this belief when he stated, remarkably, "we don't interrogate innocent people." 62 Fourth, even if an innocent suspect is somehow interrogated, police detectives tend to assume that they will realize their mistake during the questioning process 5 THOMAS GILOVICH, How WE KNow WHAT ISN'T So: THE FALLIBILITY OF HUMAN REASONING IN EVERYDAY LIFE (1991) (discussing human errors in cognition that cause them mistakenly to believe things that are not true). * Id. 56 See Keith A. Findley & Michael S. Scott, The Multiple Dimensions of Tunnel Vision in Criminal Cases, 2006 Wis. L. REV. 291 (discussing how the adversarial nature of American police interrogation institututionalizes tunnel vision). 5 See also Christian A. Meissner & Saul M. Kassin, "You're Guilty, So Just Confess!:" Cognitive and Behavioral Confirmation Biases in the Interrogation Room, in INTERROGATIONS, CONFESSIONS, AND ENTRAPMENT 85 (G. Daniel Lassiter ed., 2004). 58 Kassin & Fong, supra note 50. 5 Raymond S. Nickerson, Confirmation Bias: A Ubiquitous Phenomenon in Many Guises, 2 REV. GEN. PSYCHOL. 175 (1998) (reviewing the psychological literature on confirmation bias and demonstrating its general applicability). 60 Id. 61 Kassin & Fong, supra note 50; see also LEO, supra note 14, at 781-88. 62 Saul M. Kassin & Gisli H. Gudjonsson, The Psychology of Confessions - A Review of the Literature and Issues, 5 PSYCHOL. SCI. PUB. INT. 35, 36 (2004).

2013 WHY INTERROGATION CONTAMINATION OCCURS 205 and the interrogation will terminate. Finally, like the American public, police detectives tend to assume that innocent suspects will not confess falsely, especially to serious crimes, unless they are physically tortured or mentally ill. 64 Perhaps even more fundamentally, the failure of investigators to recognize or correct their misclassification errors is built into the structure of American police interrogation itself. Because it is guilt-presumptive by design and thus investigators believe they only interrogate suspects who are guilty, American interrogation is not and cannot be a process that sorts or separates the innocent from the guilty. It lacks a feedback mechanism to correct misclassification errors and other erroneous judgments. The goal of police interrogation is not to gather information in a neutral manner as if testing a hypothesis, but to use the psychologically forceful techniques described above to move the (presumed guilty) suspect from (his expected) denial to (the desired) admission and then to elicit a damning postadmission narrative. American police interrogation is a confirmatory, rather than an investigatory, psychological process. The objective is to use the techniques of accusatory interrogation to elicit incriminating statements that confirm the interrogators' pre-existing belief in the suspect's guilt and build a case around it. 65 As a result, police interrogators are not conditioned to hear or credit an innocent suspect's truthful denials, even when they are made loudly, emphatically and repeatedly. Rather, American police interrogators are conditioned to treat any denial(s) during an interrogation as the self-serving lies of a (presumed) guilty suspect who is seeking to escape apprehension, mislead investigators and/or minimize his culpability. Thus, investigators typically do not evaluate whether a suspect's denials, alibis or assertions of innocence are likely true, but instead assume they must be false. As a consequence, the innocent (but presumed guilty) suspect literally can neither do nor say anything to convince interrogators of his innocence, and every denial will be treated as further evidence of his guilt. But when they come from a factually innocent suspect who has been erroneously 63 See INBAU ET AL., supra note 20. 6 See Danielle E. Chojnacki et al., An Empirical Basis for the Admission ofexpert Testimony on False Confessions, 40 ARIZ. ST. L.J. 1 (2008); Mark Costanzo et al., Juror Beliefs About Police Interrogation, False Confession, and Expert Testimony, 7 J. EMPIRICAL LEGAL STUD. 231 (2010); Linda A. Henkel et al., A Survey of People's Attitudes and Beliefs About False Confessions, 26 BEHAV. Sci. L. 555 (2008); Iris Blandon-Gitlin et al., Jurors Believe Interrogation Tactics Are Not Likely to Elicit False Confessions: Will Expert Witness Testimony Inform Them Otherwise?, 17 PSYCHOL. CRIME & L. 239 (2011); Richard A. Leo & Brittany Liu, What Do Potential Jurors Know About Police Interrogation Techniques and False Confessions?, 27 BEHAV. SCI. L. 381 (2009). 65 Findley and Scott perceptively note that the very notion of American-style police interrogation "therefore, expressly embraces the foundational problems with tunnel vision-a premature conclusion of guilt, and an unwillingness to consider alternatives. In this context, however, the tunnel vision is not inadvertent, but deliberate; police are taught that this is the way to advance their investigation. Cognitive biases are openly encouraged." Findley & Scott, supra note 56, at 335 (footnotes omitted).

206 OHIO STATE JOURNAL OF CRIMNAL LAW [Vol 11: 1 misclassified as guilty, the presumed false denials, alibis and assertions of innocence are, of course, true. Nothing in the structure, sequence or psychology of American police interrogation, however, is designed to recognize, catch or correct this possibility. There is a direct link between misclassification errors and contamination in false confession cases: once there is misclassification, interrogation contamination almost inevitably follows. Misclassification begets interrogation contamination because the same process that blinds police investigators to misclassification errors also encourages interrogation contamination. The mechanism connecting the two is the confirmation biases that American police interrogation is designed to unleash into the interrogation process going forward. 6 This is particularly problematic for innocent suspects. As Saul Kassin and his colleagues have demonstrated in laboratory studies, erroneous prejudgments of guilt trigger more aggressive interrogations and cause investigators to misinterpret an innocent person's denials as the resistance of a guilty subject and redouble their efforts to elicit a confession. 6 7 Meissner and Kassin have dubbed this the "investigator bias effect." 6 8 These same processes have also been empirically observed in studies of real world interrogations leading to false confessions. 69 The tunnel vision and confirmation biases inherent in guilt-presumptive interrogation in effect create a self-reinforcing feedback loop, leading investigators to put more pressure on innocent suspects and blinding them to plausible denials, alibis and assertions of innocence, as well as to any evidence that contradicts their theory of the suspect's guilt. This observation takes us to the heart of the matter. Because American police interrogators presume the guilt of the suspects they interrogate, they also presume these suspects possess guilty knowledge (i.e., unique and non-public details) of the crime. Police investigators unwittingly educate presumed guilty (but factually innocent) suspects about non-public details of the crime because revealing inside knowledge about the crime is part and parcel of the accusatory strategies of American interrogation. These strategies are designed to break down the resistance of suspects in order to move them from denial to admission and then elicit an inculpating post-admission narrative. As part of their pre-admission interrogation strategy, investigators invariably confront a suspect with detailed case knowledge, usually through suggestive, leading and/or declarative assertions and questions, in order to expose that they know he possesses guilty knowledge as part of their effort to convince him of the futility of continued denial. As one former police investigator who elicited a 66 See Nickerson, supra note 59; Findley & Scott, supra note 56. 67 Kassin et al., supra note 22; see also Meissner & Kassin, supra note 57. 68 Meissner & Kassin, supra note 52, at 469; see also Meissner & Kassin, supra note 57. 69 See Ofshe & Leo, supra note 19.

2013 WHY INTERROGATION CONTAMINATION OCCURS 207 contaminated false confession described, "We believed so much in our suspect's guilt that we ignored all evidence to the contrary. To demonstrate the strength of our case, we showed the suspect our evidence, and unintentionally fed her details that she was able to parrot back at a later time." 70 As part of their post-admission interrogation strategy, investigators sometimes also confront a suspect with case details-again through declarative assertions, as well as leading questions, suggestions and promptings-to persuade him to stop denying his knowledge of the crime facts, to prevent him from minimizing his culpability and to fill in any missing details of the offense to which he has already confessed. 7 ' In both pre- and post-admission interrogation, American police investigators-because they believe they are interrogating only guilty individuals-do not realize when they are educating innocent individuals about specific details of how the crime occurred. Paradoxically, police interrogation contamination may be intentional, but not knowing. We now return to the basic interrogation techniques described in the previous section. As we have seen, American police investigators begin their interrogations by accusing the suspect of committing the crime, assume he will issue denials, and then accuse him of lying. The interrogators' accusations are repeated and persistent, but so too are the denials when the (presumed guilty) suspect is really factually innocent. As interrogators challenge and attack these denials as implausible, unbelievable, impossible and/or illogical, the factually innocent suspect will continue to resist and typically counter by explaining why the interrogators' assertions are mistaken. 72 In the early stages of an interrogation, an innocent suspect's denials may get stronger. For the interrogation to succeed in its goal of eliciting an admission, these denials must be countered, refuted and overcome. Interrogators often seek to counter the suspect's ongoing resistance by confronting him with detailed case information that trades on and seeks to expose his presumed guilty knowledge. By presenting the suspect with increasingly specific crime details that they believe he already knows, interrogators seek to convey that he has now been backed into a corner from which he can no longer escape, expose his denials as lies that are no longer tenable, and thus cause him to believe that his guilt has now been 73 established beyond dispute. While conveying detailed case knowledge to a guilty suspect may well lead to these perceptions and cause him to fold, it will only contaminate the knowledge base of a falsely accused innocent suspect. The psychology created by interrogators' unyielding belief in the suspect's guilt-and guilty knowledge-sets the stage for the use of evidence ploys, which is central to understanding why pre-admission interrogation contamination occurs. 70 Garrett, supra note 6, at 1075. 71 LEO, supra note 14, at 165-94. 72 See Ofshe & Leo, supra note 19. 73 LEO, supra note 14, at 139-48.

208 OHIO STATE JOURNAL OF CRIMNAL LAW [Vol 11: 1 As we have seen, an evidence ploy is any representation, whether true or false, that the interrogator either possesses or is aware of, or any type of evidence that links the suspect to the crime and/or establishes his guilt. Police typically present the real or alleged evidence against a suspect as both unequivocal and irrefutable. Related to evidence ploys are the omniscience ploys, which convey that the interrogators know all the details of the crime, as well as the suspect's alleged involvement, and therefore can discern whether he is telling the truth or lying when he responds to an accusation or question. Repeated often, evidence and omniscience ploys are among the most fundamental and effective techniques of accusatory interrogation. As with other pre-admission interrogation techniques, evidence ploys rely on the psychology of capture, exposure, and inevitability to move the suspect from denial to admission. Along with accusations and attacks on denial, evidence and omniscience ploys are used to overcome the suspect's protestations of innocence by convincing him that the objective evidence and its public revelation will, inevitably, lead to his conviction. At their most effective, evidence and omniscience ploys convince the suspect that there is no viable means of escape from his hopeless situation but to confess. 4 Police often leak and disclose non-public crime facts to innocent suspects, particularly in the pre-admission phase of interrogation, through their use of evidence and omniscience ploys. As part of their strategy to effectively challenge and reverse the suspect's denials, interrogators sometimes not only confront the suspect with invented and fabricated evidence of guilt (i.e., false evidence ploys), but also with detailed and non-public case facts (i.e., what the interrogators mistakenly believe to be true evidence ploys). This may include, for example, confronting the suspect with the existence of unique items already present at the crime scene; the order in which a set of events occurred; particular items used to commit the crime; how the crime was committed and/or the particular instrument or weapon involved in the crime; the clothing and/or other items worn by the victim; where the victim was last seen; specific things that happened to the victim before, during and/or after the crime; the names of individuals involved in the crime; and/or what the perpetrator was wearing or left at the crime scene. The most explicit forms of interrogation contamination include showing suspects police reports, witness statements, case evidence and/or crime scene photographs. Police interrogators may even include taking the suspect to the crime scene, which occurred in more than one-third of the false confession cases in the Garrett study. One of the most dramatic illustrations of pre-admission contamination in the annals of American false confession history occurred in the Phoenix Temple Four case in 1991, in which the Maricopa County Sheriffs detectives elicited four 74 id. 75 Garrett, supra note 6, at 1086.

2013 WHY INTEAROGATION CONTAMINATION OCCURS 209 proven-false confessions to the mass murder of nine Buddhist monks and their helpers. As part of their interrogation strategy in this case, Maricopa County Sheriffs detectives created a "prop room" containing crime scene photos (including pictures of the dead bodies), a diagram of the crime scene, charts, newspaper stories, and other visual evidence of what occurred during the mass murder. The interrogators brought each of the four individuals, who eventually falsely confessed, into the prop room to demonstrate law enforcement's detailed knowledge of the crime and ability to establish their guilt. Not surprisingly, after these four innocent suspects broke down, they were able to provide numerous public and non-public crime facts in their detailed, but false, confession statements. Whereas pre-admission contamination by interrogators is directed toward the goal of moving a suspect from denial to admission, post-admission contamination by interrogators is directed to helping the suspect fill in missing details and explanations in his confession narrative. Once a suspect has admitted guilt, investigators are trained to elicit information in a question and answer (as opposed to accusatory) format to allow the suspect to independently supply crime facts and explanations. Innocent suspects, however, are not likely to know non-public details of the crime that cannot be guessed by chance or that were not already disclosed by interrogators in the pre-admission portion of the interrogation. When an innocent suspect cannot provide requested crime details or incorrectly guesses crime facts that the true perpetrator must know, investigators assume he is intentionally lying and/or seeking to minimize his culpability. After all, the suspect has already admitted to the crime, thereby confirming the interrogators' belief in his guilt. At this point, investigators may easily become frustrated and impatient, thus reverting back to pre-admission interrogation techniques involving accusation, confrontation, evidence ploys, feigned omniscience and inducements to obtain the desired confession narrative. In post-admission interrogation, investigators, once again, sometimes disclose non-public crime facts to suspects through leading, suggestive and directive statements and questions whose purpose is to pressure and persuade the suspect to reveal what he supposedly knows. When innocent suspects cannot provide this information, interrogators, in frustration, sometimes tell the suspect what happened, suggest or fill in missing details, correct statements, direct him to particular conclusions, and suggest how and why the crime occurred.so 76 See GREGG 0. McCRARY, THE UNKNOWN DARKNESS: PROFILING THE PREDATORS AMONG Us 142-43 (2003); GARY L. STUART, INNOCENT UNTIL INTERROGATED: THE TRUE STORY OF THE BUDDHIST TEMPLE MASSACRE AND THE TUCSON FOUR 59, 76,242 (2010). 77 Id. 78 See LEO, supra note 14, at 165-94. 79 Leo & Ofshe, supra note 19. so Ofshe & Leo, supra note 19, at 1107.

210 OHIO STATE JOURNAL OF CRIMINAL LAW [Vol 11: 1 Once again, investigators may not be fully aware of the risk that they are contaminating the innocent since they believe they are interrogating the guilty who, by definition, already know the non-public details of the crime. As in the pre-admission portion of interrogation, investigators will try to elicit an account that is consistent with their theory of the crime. Although pre- and post-admission interrogation are directed to different short-term goals-the former to moving the suspect from denial to admission, the latter to eliciting a detailed and damning post-admission narrative-they both involve the same long-term goal of building a case against him that will ensure his conviction. As recent empirical research has demonstrated, post-admission influence and contamination may be even more robust and problematic than previously known. In addition to supplying key facts, police investigators in false confession cases also pressure, persuade, and sometimes even coerce suspects to adopt not only specific crime details, but also a narrative that creates a framework of deeper culpability. 82 This second-order level of contamination and influence during interrogation is more akin to what I have elsewhere referred to as "constructing culpability" in the "confession-making" phase of interrogation. 83 In addition to feeding or leaking details, interrogators sometimes script a suspect's postadmission narrative by suggesting how and why the crime occurred, providing possible motives and plausible explanations, correcting, suggesting and filling in missing crime-relevant information, and directing the suspect to (factual and legal) conclusions about his alleged actions and the events of the crime. 84 The police may also employ inducements to convey the benefits of giving or agreeing to a particular account, and even suggest minimizing scenarios that explain how and why the suspect committed the crime or could have done so. As a result, contaminated and scripted false confessions contain not only non-public crime facts, but a coherent and compelling story-line, motives and explanations, detailed and vivid crime knowledge, displays of emotion, description of the confessor's thoughts and feelings (both before and after supposedly committing the crime), displays of catharsis and remorse, requests for forgiveness and even expressions of voluntariness. It is this substantive content and contextual knowledge of the crime (within which non-public details are embedded) that gives the suspect's confession and narrative its full power, persuasiveness and verisimilitude. It is also why a contaminated and scripted false confession may appear to be such a seemingly compelling and convincing piece of evidence. 81 See LEO, supra note 14, at 165-94; see also Appleby et al., supra note 19, at 14-16. 82 LEO, supra note 14, at 165-94. 8 See id. 84 Id.

2013 WHYINTERROGATION CONTAMNATION OCCURS 211 IV. CONCLUSION I have argued in this article that confession contamination occurs because (1) the guilt-presumptive psychology of American police interrogation is designed to trigger and perpetuate confirmation biases, which (2) lead investigators, seemingly inadvertently, to provide detailed case information to suspects as part of their preand post-admission accusatory interrogation strategies, but (3) there is no internal corrective mechanism to catch or reverse investigators' misclassification errors or their confirmatory, information-conveying interrogation techniques. Put differently, contamination is common in false confessions because the psychological design of American interrogation methods virtually dictate the contamination once detectives have selected an innocent target for questioning. American interrogation is a laboratory of confirmation bias. Contaminated confessions are the product of what we might call the confirmation bias effect of the assumptions, goals and techniques of guilt-presumptive, accusatory police interrogation. Regardless of why it occurs, police interrogation contamination is difficult to detect. As we have seen, police investigators are trained to assiduously avoid providing suspects with non-public crime facts during interrogation; contamination is neither proper nor necessary for effective interrogation and the elicitation of reliable confessions. Thus, there is no open or legitimate justification for police interrogation contamination. To the extent it is intentional, it is an underground practice. However, if, as it appears, most police interrogation contamination is inadvertent rather than intentional-albeit a logical and predictable consequence of American methods of guilt-presumptive, accusatory interrogation-then investigators do not appear to typically realize when they are contaminating confessions. This is consistent with Garrett's finding that the detectives in his study all testified emphatically-if wrongly-in pre-trial hearings and at trial that they did not tell the suspect how the crime had occurred or supply unique crime facts through leading, suggestive or directive questioning. 85 But perhaps the most important reason why confession contamination is difficult to detect is that most interrogations are not electronically recorded. As empirical studies of proven false confessions have shown, it is rare that police have fully recorded the interrogations leading to confessions that were subsequently 86 proven false. In Garrett's study, for example, none of the interrogations in the forty cases of false confession were recorded in their entirety. 87 However, as in other studies, many of the interrogations in Garrett's data set were partially or selectively recorded, a practice that increased the risk that the interrogation 8 Garrett, supra note 6, at 1118. 86 Steven A. Drizin & Richard A. Leo, The Problem of False Confessions in the Post-DNA World, 82 N.C. L. REv. 891, 931 (2004). 87 Garrett, supra note 6, at 1079.

212 OHIO STATE JOURNAL OF CRIMINAL LAW [Vol 11: 1 contamination would go undetected by the criminal justice system. As Garrett observed about the cases in his study, "selective recording of many of these interrogations only cemented the contamination, where recording occurred after facts had already been disclosed to the innocent suspect." 88 In the absence of full recording, it is usually not possible to objectively detect confession contamination. Because it is difficult to detect, police interrogation contamination corrupts the truth-seeking process and increases the risk that a false confession will lead to a wrongful conviction. The misleading specialized knowledge contained in the suspect's post-admission narrative creates the illusion that his confession has been independently corroborated. As we have seen, these inside details are among the "content cues" that give confessions verisimilitude and cause third parties (such as prosecutors, defense attorneys, judges and jurors) to erroneously conclude in many cases that false confessions are a reliable piece of evidence on which to base a criminal conviction. 89 Police interrogation contamination not only increases the risk that a false confession will be erroneously perceived as verifiably true, but also that the criminal justice system will fail to filter it out of the stream of evidence that ultimately gets presented against a defendant. As we have seen, police and prosecutors invariably assert, in pre-trial hearings and at trial, that the unique nonpublic details contained in the confession narrative-the misleading specialized knowledge-originated with the defendant and thus establish the validity of the confession and the defendant's guilt beyond a reasonable doubt. As Garrett noted about the cases in his study: Due to the contamination of exonerees' confessions, the criminal justice system could not later untangle what transpired. Though many of these confessions displayed indicia of gross unreliability, the confessions all passed muster at trial and post-conviction... These false confessions withstood scrutiny precisely because they were bolstered by detailed facts that we now know must have been disclosed. Courts uniformly emphasized that these confessions contained admissions that only the true murderer or rapist could have known. Perhaps not surprisingly, several empirical studies have shown that the overwhelming majority-73% to 8 8 /- of proven false confessors who take their case to trial before their innocence has been proven are erroneously convicted. 90 These figures are even higher if we include false confessions leading to 8 Id. at 1118. 89 LEo, supra note 14, at 165-94. 90 See Leo & Ofshe, supra note 19, at 484; Drizin & Leo, supra note 86, at 931; Jon Gould et al., Predicting Erroneous Convictions, 99 IOWA L. REV. (forthcoming 2014).

2013 WHY INTERROGATION CONTAMINATION OCCURS 213 convictions by plea bargain. 9 ' These findings are consistent with experimental research on the impact and potency of confessions on triers of fact, 92 as well as conventional legal wisdom about the uniquely prejudicial nature of confession evidence. 93 It is only possible to objectively determine whether police investigators have contaminated the confessor's narrative if the interrogation is recorded in its entirety. As of this writing, sixteen states and the District of Columbia now require electronic recording of custodial interrogations as a matter of law in at least some cases. 94 As a policy reform, the mandatory electronic recording of custodial interrogations would ensure that interrogation contamination is captured on tape and reviewable by outsiders. As Garrett argues, "to complete interrogation record enables meaningful reliability review and could help prevent the problem of confession contamination." 95 Though valuable for a number of reasons, 96 a mandatory recording requirement, however, is not a perfect solution to the problem of police interrogation contamination. If interrogation contamination really is inadvertent, a mandatory electronic recording requirement by itself may not prevent interrogation contamination so much as reduce the number of erroneous convictions based on contaminated false confession evidence. In addition, because it does not appear feasible to capture all police-suspect encounters on tape, in at least some cases police investigators could leak non-public crime facts to suspects off-tape, thereby preventing criminal justice officials and others from objectively determining the source from which these details originated. As Garrett observes: even a mandatory recording regime can be circumvented, perhaps with ease... if police do occasionally depart from training and procedure, and try to disclose key facts to a suspect then they may also try to circumvent a requirement that custodial interrogations be recorded. After all, police may have contact with a suspect outside of the interrogation room, in 91 See Ofshe & Leo, supra note 19. 92 Saul M. Kassin, Why Confessions Trump Innocence, 67 AM. PSYCHOLOGIST 431, 434 (2012). 9 As the United States Supreme Court has noted, "no other class of evidence is so profoundly prejudicial... [t]riers of fact accord confessions such heavy weight in their determinations that 'the introduction of a confession makes the other aspects of a trial in court superfluous, and the real trial, for all practical purposes, occurs when the confession is obtained."' Colorado v. Connelly, 497 U.S. 157, 182 (1986). 94 See Thomas S. Sullivan, A Compendium of Law Relating to the Electronic Recording of Interrogations, 95 JUDICATURE 212, 213-14 (2012). The sixteen states are: Alaska, Minnesota, Illinois, Wisconsin, Maine, New Mexico, New Jersey, North Carolina, Maryland, Montana, Missouri, Oregon, Indiana, Connecticut, Nebraska, and Michigan. 95 Garrett, supra note 6, at 1113. 96 See LEO, supra note 14, at 291-305.